AdverSparse: An Adversarial Attack Framework for Deep Spatial-Temporal Graph Neural Networks

Abstract

Spatial-temporal graph have been widely observed in various domains such as neuroscience, climate research, and transportation engineering. The state-of-the-art models of spatial-temporal graphs rely on Graph Neural Networks (GNNs) to obtain explicit representations for such networks and to discover hidden spatial dependencies in them, leading to superior performance in various tasks. In this paper, we propose a sparse adversarial attack framework AdverSparse to illustrate that when only a few key connections are removed in such graphs, hidden spatial dependencies learned by such spatial-temporal models are significantly impacted, leading to various issues such as increasing prediction errors. We formulate the adversarial attack on such models as an optimization problem and solve it by the Alternating Direction Method of Multipliers (ADMM). Experiments show that AdverSparse can find and remove key connections in these graphs, leading to malfunctioning models, even in models capable of learning hidden spatial dependencies.

BibTeX Copy

@inproceedings{jiayu2022adversparse,
  title = {AdverSparse: An Adversarial Attack Framework for Deep Spatial-Temporal Graph Neural Networks},
  author = {Jiayu Li and Tianyun Zhang and Shengmin Jin and Makan Fardad and Reza Zafarani},
  year = {2022},
  keywords = {conference},
  booktitle = {Proceedings of the 2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
  abstract = {Spatial-temporal graph have been widely observed in various domains such as neuroscience, climate research, and transportation engineering. The state-of-the-art models of spatial-temporal graphs rely on Graph Neural Networks (GNNs) to obtain explicit representations for such networks and to discover hidden spatial dependencies in them, leading to superior performance in various tasks. In this paper, we propose a sparse adversarial attack framework AdverSparse to illustrate that when only a few key connections are removed in such graphs, hidden spatial dependencies learned by such spatial-temporal models are significantly impacted, leading to various issues such as increasing prediction errors. We formulate the adversarial attack on such models as an optimization problem and solve it by the Alternating Direction Method of Multipliers (ADMM). Experiments show that AdverSparse can find and remove key connections in these graphs, leading to malfunctioning models, even in models capable of learning hidden spatial dependencies.},
  pdf = {files/AdverSparse.pdf},
}